Assessing the Robustness of Web Feature Services Necessary to Satisfy the Requirements of Coastal Management Applications by Jennifer R. Bauer A Research Paper submitted to The College of Earth, Ocean, and Atmospheric Sciences, Oregon State University in partial fulfillment of the requirement for the degree of Master of Science Geography Program June, 2012 Committee in Charge: Dawn Wright, Geosciences Julia Jones, Geosciences David Hart, University of Wisconsin Sea Grant Institute Table of Contents Acknowledgements Abstract Introduction………………………………………………………………………. 1 Methodology……………………………………………………………………... 7 Evaluating Web Feature Service Components……………………………….. 8 Software………………………………………………………………….. 8 Web Mapping Servers……………………………………………….. 11 Desktop GIS Applications……………………………………………. 13 Hardware…………………………………………………………………. 14 Data Characteristics………………………………………………………. 14 Number of Features………………………………………………...… 14 Feature Attributes…………………………………………………….. 15 Metadata……………………………………………………………… 16 Distributed Spatial Queries…………………………………………………... 17 Results……………………………………………………………………………. 18 Software……………………………………………………………………… 18 Hardware…………………………………………………………………...... 21 Data Characteristics………………………………………………………...... 23 Distributed Spatial Queries………………………………………………….. 24 Discussion………………………………………………………………………... 25 Conclusions………………………………………………………………………. 31 References……………………………………………………………………...... 33 List of Figures Figure Page 1. Home page of the Wisconsin Coastal Atlas………………………………….. 6 2. Example of a basic client-server architecture (A) and a client-server architecture for a WFS (B)………………………………….... 9 3. Example of a <get capabilities> WFS response from MapServer ………….. 10 4. Study area of Douglas, Bayfield, Ashland, and Iron Counties in Wisconsin ………………………………………………….. 12 5. Comparison of average download time (s) and file size (Mb) for WFS requests to MapServer, GeoServer, and ArcServer from a Python script (n=30)……………………………………………………….. 19 6. Comparison of average download time (s) and file size (Mb) for WFS requests from Quantum GIS, gvSIG, and ArcGIS (n=10)…………………... 21 7. Comparison of average WFS download time (s) and file size (Mb) across different network capabilities………………………………………… 22 8. Average download time (s) for <get feature> WFS requests created with various data attribute quantities………………………………… 23 9. Example of a MapServer Map File…………………………………………… 26 List of Tables Table Page 1. Comparison of three popular OGC web services…………………………….. 5 2. Comparison of MapServer, GeoServer, and ArcServer……………………... 11 3. Comparison of Quantum GIS, gvSIG, and ArcGIS………………………….. 13 4. Number of features included in each county’s WFS…………………………. 15 5. Metadata components included in each WFS for Douglas County….…….... 17 6. Average file size (Mb) and download time (s) for WFS’s distributed by MapServer, GeoServer, and ArcServer….……………………………….. 18 7. Simple linear regression results for average WFS download time (s) and file size (Mb) from all three web mapping servers……………………… 19 8. Average file size (Mb) and download time (s) for WFS accessed with Quantum GIS, gvSIG, and ArcGIS..……………………………………. 20 9. Regression analysis results for average WFS download time (s) and file size (Mb) from all three desktop GIS applications……………………… 21 10. Average WFS download time (s) and file size (Mb) from MapServer, GeoServer, and ArcServer for WFS’s created with various levels of metadata...…………..………………………………… 24 11. Spatial query results utilizing WFSs to calculate the economic value of coastal properties along the Lake Superior coast of Wisconsin..……………. 25 Acknowledgements I would like to thank several individuals that have contributed to the development of this research project and to my success at Oregon State University. Dr. Dawn Wright, my major advisor, was instrumental in providing assistance with developing this research project, and I would like to thank her for all of her patience, unwavering support, and guidance during this process. A special thanks to Dr. David Hart of the University of Wisconsin-Madison, one of my committee members, for his encouragement, enthusiasm, and support throughout the development of this research and for providing critical background information about the Wisconsin Coastal Atlas and web feature services. I would also like to thank my other committee member, Dr. Julia Jones for her unconditional support, her creative input that helped spark new ways to approach problems and consider solution, as well as her guidance during the final year of my tenure here at Oregon State. I wish to extend special thanks to Dr. James Graham for all his support, web-based GIS and python expertise, and interest in my research as well. I would also like to extend my gratitude to the entire facility and staff of the Geosciences department and my colleagues for supporting and assisting me during my time here, particularly my fellow rogues of Davey Jones Locker, Kelvin Raiford, Colleen Sullivan, and Dori Dick, for their help and compassion. A huge thank you goes out to all my family and friends scattered across the U.S. for unwavering support and constant encouragement for me throughout this entire process. Special thanks to my parents, Charles and Linda Olson for instilling the importance of education and always encouraging me to follow my dreams. Finally, I would like to thank my wonderful husband, David, for his love and support during this entire process, from the GRE to graduate school applications, moving from Texas to Oregon, and for doing everything in his power to help me succeed here at Oregon State. Thank you! Assessing the Robustness of Web Feature Services Necessary to Satisfy the Requirements of Coastal Management Applications Abstract Ever expanding pressures on the health and productivity of our oceans and coasts from threats such as coastal development and climate change are stressing the need to consider the full spectrum of factors, scales, datasets, opinions, and trade-offs for current and future coastal management actions (Guerry 2009; McLeod and Leslie 2009; Rosenberg and Sandifer 2009). Web-based GIS tools including coastal web atlases (CWAs) and geospatial web services are being rapidly developed to assist managers, decision-makers, and scientists with the creation, implementation, and evaluation of coastal management options. Numerous CWAs are incorporating web feature service (WFS) standards into their websites to provide critical datasets on-the-fly for use in spatial decision support tools to assist with management and policy decisions. However, with this increased use of WFSs in CWAs it’s critical to understand how the various components used to create robust WFSs can affect its performance and ability to successfully execute complex spatial queries that are utilized to assess management options. A subset of county land parcel data from Wisconsin was utilized to assess how various software, hardware, and data characteristics can affect WFSs overall robustness, and how these components impact its ability to execute accurate, timely complex spatial queries consistently and their ability to meet the demands of managers, decision-makers, and scientists. Results suggest that WFSs, with varying levels of robustness, can successfully perform accurate, reliable spatial queries on datasets to extract relevant information pertaining to coastal management concerns, which could impact CWAs by increasing their functionality and range of potential applications provided to users. This research paper follows the style guidelines of the Annals of the Association of American Geographers 1 Introduction The severity and scale of threats challenging the health and productivity of coastal ecosystems are increasing at an unprecedented rate (e.g., McLeod and Leslie 2009). The range of pressures faced by these ecosystems are diverse, including threats from point and nonpoint source pollution, habitat loss and fragmentation, climate change, ocean acidification, sea level rise, shoreline erosion, and coastal development (McLeod et al. 2005; Lubchenco 2009). Research has shown that the impacts from these threats are often widespread and adverse (Feeley et al. 2008; Lubchenco 2009; McLeod and Leslie 2009; FAO 2010). Examples include estimates that 85% of the world’s fish stocks are at or above maximum sustainable yield, over 20,000 acres of critical habitat including wetlands and mangroves disappear annually, and over 150 invasive species have been introduced to U.S. waters since 1970 (Pew Ocean Commission 2003; FAO 2010). However, impacts from these threats aren’t limited solely to coastal ecosystems, but also affect human society and culture. The coasts are home to 50% of the global population, with suggestive estimates that by 2020 approximately 75% of the global population will live within 60km of the coast, and over a billion people rely on fisheries as their major source of protein in their diet (Feeley et al. 2008; FAO 2010; McGlade 2011; Wright, Dwyer, and Cummins 2011). Humans are also reliant upon other services these ecosystems provide including transportation, water filtration, renewable energy, carbon sequestration, erosion control, and various recreational usages (e.g., McLeod and Leslie 2009). This intrinsic complexity between humans and coastal ecosystems, in combination with numerous threats they face, pose a unique challenge to managers, decision-makers, and scientists tasked with developing new, effective management strategies. Traditional management strategies, which focus on individual sectors of